318 
APPENDIX II. 
a nd points out that the larger the atoll the purer the sea- 
water of the lagoon, so that the latter would speedily reach 
a limit to its expansion in consequence of the non-growth 
of coral; indeed, as a matter of fact, these larger lagoons 
contain plenty of living coral. The second (and ultimately 
more important) factor in the enlargement of a lagoon, 
viz. the corrosion of the dead coral by the solvent action 
of water, he regards as a hypothesis which has little direct 
evidence in its favour and much indirect against it. The 
connection of channels with prevailing currents cannot be 
shown, and the former in many cases are sufficiently deep 
to be well below the limit of abrasion. Indeed, as a rule, 
so far from these channels being enlarged by solution and 
abrasion, they tend to be closed by the growth of living 
coral, and many of the lagoons in the smaller islands are 
without channels. Hence in them, as there can be no 
appreciable transference of water, the action of solution 
must be reduced to a minimum. Yet these closed atolls 
are by no means exceptional. For instance, in the case of 
about sixty coral islands mapped by the Wilkes expedition, 
of those which range from 1^ to 3 miles in the longer 
diameter of the reef, nine have no lagoon, but only a small 
depression in its place, which is dry, except in the case of 
two where water gets in at high tide. Of those under 6 
miles in length, having lagoons, 17 in number : 16 have 
no entrance to the lagoon at low tide ; the other has an 
entrance of large size. Of those 6 miles or over in length, 
29 in number : 17 have channels and 12 have none; those 
having channels are generally over 10 miles long. It 
must also be understood that the opponents of the subsid- 
ence theory are compelled to admit it in order to explain 
the depth of certain lagoons. Hence Professor Dana con- 
siders that the hypothesis of elevation or lateral spreading 
during a period of rest is inadequate as a general expla- 
nation of the problem. 
